We characterized the spatiotemporal epidemiology of rabies from January 2009 through March 2014 across the interface between a wildlife reserve and communal livestock farming area in South Africa. Brain tissue from 344 animals of 28 different species were tested for lyssavirus antigen. Of these, 146 (42.4%) samples tested positive, of which 141 (96.6%) came from dogs. Brain samples of dogs were more likely to test positive for lyssavirus antigen if they were found and destroyed in the reserve, compared to samples originating from dogs outside the reserve (65.3% vs. 45.5%; odds ratio (OR) = 2.26, 95% confidence interval (CI) = 1.27-4.03), despite rabies surveillance outside the reserve being targeted to dogs that have a higher index of suspicion due to clinical or epidemiological evidence of infection. In the reserve, dogs were more likely to test positive for rabies if they were shot further from villages (OR = 1.42, 95% CI 1.18-1.71) and closer to water points (OR = 0.41, 95% CI 0.21-0.81). Our results provide a basis for refinement of existing surveillance and control programs to mitigate the threat of spillover of rabies to wildlife populations.